Between-hospital variability in organ donation after resuscitation from out-of-hospital cardiac arrest

Abstract

Introduction:

Survival and recovery after out-of-hospital cardiac arrest (OHCA) varies between hospitals, with better outcomes associated with high-volume and specialty care. We evaluated if there is a similar relationship with organ donation after OHCA.

Methods:

We studied a cohort of adults resuscitated from OHCA from 2010 to 2018, treated at one of 112 hospitals served by a regional organ procurement organization (OPO). We obtained hospital-level characteristics from Centers for Medicare and Medicaid Services and Health Resources and Services Administration and obtained patients’ clinical information from the OPO health record. We excluded patients with no potential to donate on initial referral. Our primary exposure was treatment at a high-volume hospital (defined >500 eligible cases during the study period) and our primary outcomes were suitability to donate after full medical evaluation, successful organ procurement and organ transplantation. We used mixed effects models to quantify between-hospital variability in the primary outcomes.

Results:

Overall, 9,792 patients were included and 796 (8%) were organ donors. We identified significant between-hospital variation in odds of donation (median odds ratio 1.64 [95% CI 1.42 – 2.02]). Hospital volume explained the greatest proportion of variability. High volume centers had a higher proportion of referrals with potential to donate (16.9 vs 12.2%), actual donation (10.3 vs 6.2%), and successful transplantation (9.4 vs 5.7%). Overall, 2032/7763 (26%) of recovered transplantable organs in this region were procured from OHCA patients.

Conclusion:

High volume centers are more likely to refer and procure transplantable organs from patients with non-survivable OHCA.

Introduction

Wait times for organ transplantation continue to increase in the United States because the demand for donated organs exceeds supply.¹ There is considerable variation in the frequency with which imminently dying potential donors engage with organ procurement organizations (OPOs) and successfully donate.^1,2 Patient and family factors account for only a portion of the observed variability. Hospital-level factors may also contribute, but little is known about the magnitude of between-hospital variability or hospital characteristics associated with successful donation.

Tiered hospital accreditation and regionalization systems concentrate care of patients presenting with brain trauma or severe stroke at specialty centers.^3,4 Despite growing evidence that patients resuscitated from cardiac arrest also benefit from specialty care,^5,6 regional systems of post-arrest care are not widely implemented. These patients therefore may be particularly prone to experience variation in quality of care and outcome based on differences attributable to their treating hospitals. As many will not survive, post-cardiac arrest patients may also comprise a substantial proportion of organ donors.⁷

We previously described substantial improvement in long-term survival when post-arrest patients receive specialty care at high volume receiving centers, suggesting an opportunity to improve patient recovery through quality improvement and/or regionalization of cardiac arrest care.⁵ In this study, we explore whether there is similar between-hospital variability in rates of successful organ donation by patients who do not recover from out-of-hospital cardiac arrest (OHCA). Such variability would suggest that there is a parallel opportunity to improve organ availability through quality improvement efforts and/or regionalization of OHCA care. We had two main objectives: (1) determine if variation in organ donation after resuscitation from OHCA is attributable to both patient- and hospital-level characteristics, and (2) determine the proportion of organ donors in the region referred from patients who had OHCA.

Methods

Patients and Setting

The University of Pittsburgh Office of Human Research Protection (formerly Institutional Review Board) approved this project. We performed a retrospective cohort study of patients resuscitated from OHCA between January 2010 and March 2018 who were cared for at one of 113 acute care hospitals in regions of New York, western Pennsylvania and West Virginia served by the Center for Organ Recovery and Education (CORE). We excluded patients <14 years of age, those already deceased at the time of referral, and those who survived to hospital discharge. We further excluded those who met preliminary screening criteria as having no potential to donate organs because they were not mechanically ventilated, had known active malignancy, abdominal sepsis, or positive human immunodeficiency virus serostatus (prior to passage of the HIV Organ Policy Equity Act). Finally, we excluded patients referred from hospitals with fewer than one referral per year over the course of the study period.

CORE is one of 58 federally designated OPOs responsible for coordinating deceased organ donation in the United States and serves a catchment area of approximately 5.5 million. Under federal regulations, all decedents and patients with imminent death must be referred to their local OPO for consideration of organ and tissue donation. CORE staff review hospital mortality records on an ongoing basis to ensure compliance with this requirement. CORE operates a single call center with staff who use a structured set of questions to elicit specific aspects of referred patients’ history (including the inciting event leading to hospitalization and referral), vital signs, laboratory results, key exam findings, and plan of care at the time of referral. Call center staff immediately contact a clinical organ procurement coordinator (OPC) on call when potential organ donors meeting preliminary screening criteria are identified. In these cases, the OPC works with hospital staff to collect and record additional detailed clinical information necessary to assess potential for donation. For patients deemed to have potential to donate organs after determination of death by neurological criteria (brain dead donors (BDD) or circulatory criteria (donors after cardiac death (DCD)), the OPC approaches family to seek surrogate authorization or honor first-person authorization for donation. All DCD cases in our region occur after planned withdrawal of life-sustaining therapies (i.e., Maastricht Category III).

We developed and refined a text-based search of CORE referrals to identify patients referred after cardiac arrest. We iteratively reviewed random samples of 500 returned results and adapted the query strategy to reduce inclusion of referrals for conditions other than cardiac arrest, until our search query demonstrated 95% specificity. To further improve our search accuracy, we also compared returned results to a prospective registry of post-arrest patients treated at the highest volume referral center in the region,⁵ identified missed cases, and refined the query to reach 95% sensitivity. To minimize inclusion of patients resuscitated from in-hospital cardiac arrest, we restricted results to patients referred within one calendar day of admission.

Hospital-level predictors

Our primary exposure was the treating hospital status from which each patient was referred to CORE, which we categorized as high volume if treated >500 OHCA cases meeting inclusion and exclusion criteria during our study period.⁵ We obtained hospital-level characteristics from Centers for Medicare and Medicaid Services (CMS) annual Healthcare Cost Report Information System (HCRIS) data. Hospital characteristics included academic status (based on resident-to-bed ratio, defined as small teaching, large teaching, or non-teaching hospital), number of licensed in-patient beds; number of licensed intensive care unit (ICU) beds, and hospital financial model (defined as for-profit, not-for-profit, or government facility). We used Health Resources and Services Administration data to identify hospitals operating in Medically Underserved Areas, regions designated based on the availability of primary care providers. We used data from CORE to identify transplant centers and hospitals recognized for excellence in organ donation (e.g. the West Virginia Governor’s Award for Life, the Pennsylvania Department of Health’s Donate Life Pennsylvania Hospital Challenge, etc).

Patient-level predictors and outcomes

We obtained clinical information from CORE’s electronic health record, including referred patients’ age; sex; systolic and diastolic blood pressure at the time of referral; vasoactive infusions, from which we specifically considered as separate binary predictors administration of dopamine, norepinephrine, epinephrine, phenylephrine, vasopressin and inotropes (milrinone or dobutamine); laboratory values (aspartate transaminase (AST), alanine transaminase (ALT), alkaline phosphatase, total bilirubin, blood urea nitrogen (BUN), creatinine); urine output, which we categorized as none or oliguric (<0.5mL/kg/hr or <35mL/hr if weight was unknown), non-oliguric, or unknown (for example, in patients referred immediately upon hospital arrival and before Foley catheter placement); neurological exam findings, recorded as presence or absence of pupillary light reflex, corneal reflex, cough, gag reflex, respiratory drive over the set rate of mechanical ventilation, response to pain, and administration of any sedation that might confound neurological assessment; time of referral and time of death.

We identified the total number and types of organs procured, total number and types of organs transplanted (excluding organs that were procured and discarded or dispositioned to research), types of donor (DCD vs BDD, with differentiation of standard criteria and expanded criteria BDDs according to United Network of Organ Sharing (UNOS) criteria) directly from the CORE record.⁸

We used three temporally related outcomes for our main regression models. First, we assessed potential for organ donation, which we defined as present when an OPC approached family to discuss donation. Second, we considered conversion to an organ donor, which we defined as procurement of at least one organ with the intent for transplantation. Finally, we considered successful transplantation of at least one procured organ into a recipient.

Statistical analysis

We used descriptive statistics to summarize patient and hospital characteristics. We used multiple imputation with chained equations to impute missing patient-level. We determined the necessary number of imputations by estimating Monte Carlo errors in analyses of pooled results.⁹ For each patient-level outcome, we used hierarchical logistic regression models to quantify between-hospital variability, which we summarized as median odds ratios.¹⁰ First, we constructed a model with no patient- or hospital-level covariates to quantify crude between-hospital variability. We then determined the percent of variation explained by constructing nested models sequentially adding each hospital-level characteristic as a random slope and compared the residual between-hospital variability to the base model. Substantial multicollinearity precluded simultaneous adjustment for multiple hospital-level characteristics. Next, we adjusted for patient-level characteristics and finally adjusted for both patient- and the single most influential hospital factor. Finally, as a post hoc analysis we used locally weighted scatterpot smoothing (LOWESS) to create plots describing the relationship between hospital volume and outcomes. We performed data cleaning using R (R Foundation for Statistical Computing, Vienna, Austria) and performed multiple imputation and regression analyses using Stata Version 15 (StataCorp, College Station, TX).

Results

We identified 49,281 patients referred to CORE within one calendar day of hospitalization following OHCA of whom 9,792 were included in analysis (Figure 1). The most common reason for exclusion was posthumous referral for consideration of tissue donation only. Included patients were cared for at one of 54 acute care hospitals with median 74 [interquartile range (IQR) 24 – 195] patients referred per hospital (Table 1), with 4,317 cared for at one of 5 high-volume centers. Mean age was 62 ± 17 years and 4,451 (45%) were female (Table 2). Overall, 1,409 patients (1,029 brain dead (11% of the included cohort); 380 non-brain dead) were determined to have potential as organ donors after full clinical evaluation and had family approached by an OPC to discuss donation. Of these, 949 (67%) authorized donation, without significant between-hospital variation in authorization rate (P = 0.28). Of authorized cases, 796 (84%) converted to organ donors (176 DCD donors and 620 BDD (482 standard criteria donors and 138 extended criteria donors)) and 728 (91% of donors) had at least one procured organ transplanted into a recipient (153 DCD donors and 575 BDD (458 standard criteria donors and 117 extended criteria donors)). High volume centers had a higher proportion of referrals with potential to donate (16.9 vs 12.2%, P <0.001), actual donation (10.3 vs 6.2%, P <0.001), and successful transplantation (9.4 vs 5.7%, P <0.001). Each of these outcomes showed a quasi-linear relationship with hospital outcome when plots were fit using LOWESS (Figure 2).

Figure 1:

STROBE diagram

Table 1:

Characteristics of included hospitals

Hospital characteristic	Overall (n = 54)
Cardiac arrest referrals	74 [24 – 195]
<25	13 (24)
25 - 99	18 (33)
100 - 199	9 (17)
200 - 499	9 (17)
≥500	5 (9)
Licensed beds	198 [132 – 286]
<100	5 (9)
100 – 199	22 (41)
200 – 399	20 (37)
≥400	7 (13)
Teaching hospital	30 (56)
Transplant center	3 (6)
Proprietary corporation (vs nonprofit)	11 (20)
Medically underserved area	21 (39)
Received state donation award	28 (52)

Data are presented as median [interquartile range] or number with corresponding percentage.

Table 2:

Patients’ clinical characteristics at time of referral

Characteristic	Overall cohort (n = 9,792)	Potential as donor (n = 1,409)	Organ donor (n = 796)	Successful transplantation (n = 728)
Age, years	62 ± 17	45 ± 15	42 ± 21	41 ± 14
Female sex	4,451 (45)	639 (45)	363 (46)	331 (45)
Transferred	264 (3)	49 (3)	31 (4)	29 (4)
Hemodynamics
Systolic blood pressure, mmHg	114 ± 32	123 ± 31	123 ± 30	123 ± 30
Diastolic blood pressure, mmHg	63 ± 19	72 ± 21	74 ± 21	74 ± 21
Epinephrine infusion	1,967 (20)	179 (13)	97 (12)	87 (12)
Norepinephrine infusion	4,067 (42)	448 (32)	249 (31)	231 (32)
Vasopressin infusion	1,522 (16)	102 (7)	60 (7)	56 (8)
Phenylephrine infusion	695 (7)	51 (4)	27 (3)	25 (3)
Dopamine infusion	929 (9)	102 (7)	49 (7)	44 (6)
Inotrope infusion	81 (1)	2 (0)	1 (0)	1 (0)
Laboratory values
AST	138 [58 – 397]	153 [76 – 329]	155 [76 – 335]	156 [79 – 338]
ALT	84 [37 – 238]	111 [56 – 231]	113 [58 – 237]	117 [63 – 242]
Total bilirubin	0.6 [0.4 – 1.1]	0.4 [0.3 – 0.7]	0.4 [0.3 – 0.7]	0.4 [0.3 – 0.7]
Alkaline phosphatase	95 [71 – 135]	95 [74 – 125]	95 [74 – 119]	94 [75 – 120]
Blood urea nitrogen	24 [16 – 39]	16 [12 – 22]	15 [11 – 20]	15 [11 – 19]
Creatinine	1.6 [1.1 – 2.5]	1.3 [1.0 – 1.7]	1.2 [0.9 – 1.6]	1.2 [0.9 – 1.6]
Urine output
Unknown	4,225 (43)	608 (43)	347 (44)	317 (44)
Anuric or oliguric	3,091 (32)	233 (17)	99 (12)	89 (12)
Adequate	4,276 (25)	568 (40)	350 (44)	322 (44)
Neurological exam findings
Pupillary light reflex	3,000 (31)	239 (17)	137 (13)	125 (17)
Corneal reflex	2,255 (23)	153 (11)	83 (10)	77 (11)
Gag reflex	2259 (23)	149 (10)	86 (11)	78 (11)
Cough reflex	2585 (26)	202 (14)	117 (15)	106 (15)
Responds to pain	1,921 (20)	128 (9)	75 (9)	67 (9)
Pharmacologically sedated	3,147 (32)	406 (30)	240 (30)	217 (30)

Figure 2:

LOWESS smoothing revealed a quasi-linear relationship with increased proportion of referred OHCA patients deemed to have potential to donate (A), actual donation (B), and procurement of >=1 organ that was successfully transplanted (C).

We performed 20 imputations prior to regression analysis. We identified significant between-hospital variation in the odds a patient was deemed to have potential to donate (median odds ratio (MOR) 1.54 [95% CI 1.36 – 1.82], P value <0.001 for likelihood ratio test versus a fixed effects model), those who donated (MOR 1.64 [95% CI 1.42 – 2.02], P<0.001), and odds of donation with subsequent transplantation (MOR 1.67 [95% CI 1.43 – 2.10], P<0.001) (Table 3). Across outcomes, number of licensed beds explained the greatest proportion of between-hospital variability (51 to 64% of total variance). By contrast, teaching status, hospital control type (proprietary vs nonprofit) and location in a medically underserved area each explained little variability. Multiple patient-level characteristics were associated with outcome. Adjusting for these explained 62 to 74% of total variance between hospitals (Table 3). Adjustment for patient factors and hospital size (number of licensed beds) explained 74% of variance between hospitals in potential to donate at the time of referral (MOR 1.20 [95% CI 1.06 – 1.35]). Insufficient outcomes at many hospitals precluded simultaneous adjustment for patient- and hospital characteristics in models predicting organ donation and successful transplantation.

Table 3:

Median odds ratios between randomly selected hospitals quantify unexplained between-hospital variation.

Adjusted model	Potential to donate		Organ donation		Successful transplantation
	Median odds ratio [95% CI]	Variance explained, %	Median odds ratio [95% CI]	Variance explained, %	Median odds ratio [95% CI]	Variance explained, %
Unadjusted	1.54 [1.36 – 1.82]	Ref	1.64 [1.42 – 2.02]	Ref	1.67 [1.43 – 2.10]	Ref
Licensed beds	1.35 [1.21 – 1.60]	51	1.35 [1.19 – 1.68]	64	1.38 [1.20 – 1.76]	60
Transplant center	1.39 [1.25 – 1.64]	40	1.45 [1.27 – 1.77]	43	1.48 [1.29 – 1.85]	41
High-volume center	1.47 [1.31 – 1.72]	21	1.49 [1.30 – 1.82]	36	1.51 [1.31 – 1.88]	36
State donation award	1.49 [1.32 – 1.75]	15	1.58 [1.36 – 1.95]	15	1.60 [1.37 – 2.01]	16
Teaching status	1.52 [1.35 – 1.80]	4	1.61 [1.39 – 1.98]	9	1.62 [1.39 – 2.04]	11
Proprietary corporation	1.53 [1.35 – 1.81]	3	1.60 [1.39 – 1.96]	10	1.63 [1.40 – 2.04]	8
Medically underserved area	1.52 [1.35 – 1.81]	2	1.63 [1.41 – 2.02]	2	1.67 [1.43 – 2.09]	1
Patient factors only	1.24 [1.11 – 1.39]	62	1.22 [1.06 – 1.40]	76	1.20 [1.03 – 1.40]	81
Patient and licensed beds	1.20 [1.06 – 1.35]	74	--	--	--	--

Overall, 7763 recovered organs procured in CORE’s catchment area during the study period were successfully transplanted, of which 2032 (26%) were recovered from patients referred after resuscitation from OHCA (Table 4). Mean organs transplanted per DCD OHCA donor was 1.9 ± 1.0, mean organs transplanted per standard criteria brain dead OHCA donor was 3.1 ± 1.6, and mean organs transplanted per extended criteria brain dead OHCA donor was 1.6 ± 1.2.

Table 4:

Organs transplanted from patients resuscitated from out-of-hospital cardiac arrest as the proportion of total organs transplanted regionally during the study period.

Organ	Number transplanted
Kidney	1081/3112 (35)
Liver	536/1753 (31)
Lung	200/1843 (11)
Heart	170/831 (20)
Pancreas	45/224 (20)
Total organs	2032/7763 (26)

Data are presented as: Numerator: total organs of each type procured and transplanted from OHCA patients meeting inclusion criteria for this study; Denominator: overall total organs of each type procured and transplanted in the region regardless of OHCA status; corresponding percentage.

Discussion

We identified substantial between-hospital variability in donation-related performance metrics, with significantly better results at larger centers and those centers that treated a high volume of potential OHCA donors. This finding parallels volume-outcome relationships observed in many acute care conditions,¹¹ and complements a growing body of research that shows improved survival and favorable recovery when OHCA patients are cared for at specialty or high-volume centers.^5,12-18 In our cohort, patients hospitalized after resuscitation from OHCA provided 26% of the overall organs that were successfully transplanted. Importantly, prior research has shown recipients’ outcomes are comparable when organs were procured from patients resuscitated from cardiac arrest versus other donors.^19,20 Taken together, these observations support the concept that regionalized post-arrest care can both save lives and substantially increase availability of organs available for transplantation.

In high-income nations like the United States and United Kingdom, OHCA patients represent a growing proportion of organ donors²¹ and organ donation is widely viewed as an important benefit to resuscitation.^20,22 Beyond arguments for regionalization, potential to increase availability of organs for transplantation shapes consensus recommendations and interpretation of clinical research. For example, improved short-term survival leading to opportunities for donation is considered an important potential benefit of intra-arrest epinephrine administration, despite data suggesting no benefit to 30-day functionally favorable survival.^23,24

To estimate the potential effect of regionalized post-arrest care on organ availability, we extrapolate from our present findings and epidemiological estimates of OHCA. The proportion of additional organs procured at high-volume centers parallels the improvement in improved survival we previously demonstrated in an overlapping region.⁵ In this prior cohort, 30-day survival at high-volume receiving centers was 41% (vs 27% elsewhere) and hospital characteristics of these receiving centers were comparable to the high volume donation centers identified in the present analysis. Assuming 89,000 patients survive to hospital care after OHCA in the United States annually,²⁵ our prior and current data suggest transferring just half of the patients currently cared for at low volume centers to specialty care could result in both >5,000 OHCA lives saved annually and >2,000 additional lives saved through transplantation.

Data available in our study were insufficient to determine the reason for improved process-related outcomes at high-volume centers. Measurable patient characteristics explained a large proportion of between-hospital variability. This is not unexpected, as hospital care prior to the time of referral, such as aggressiveness of initial resuscitation, is likely also to affect suitability of potential donors at the time of referral. Processes of care associated with improved survival after OHCA such as coronary revascularization, aggressive hemodynamic resuscitation and delayed withdrawal of life-sustaining therapy also improve end-organ perfusion or allow time for early multisystem organ failure to recover.^20,26-28 Thus, higher quality life-sustaining efforts prior to death preserve the possibility of organ donation among decedents. Conversely, therapeutic nihilism may both worsen survival and adversely affect potential to donate.^27,29

Our study has limitations. While many baseline characteristics were available for analysis, there are likely other unmeasured sources of between-patient and between-hospital heterogeneity. For example, since CORE’s database is not specific to the cardiac arrest population, standard Utstein-style data (e.g. presenting arrest rhythm, arrest etiology, etc.) are not systematically collected at the time of referral and were thus unavailable for analysis. With regard to hospital-level predictors, considerable multicollinearity prevented us from jointly adjusting for multiple characteristics as random coefficients. As a practical matter, however, there are few high-volume referral centers that are also non-academic or low-volume for other procedures, therefore the multicollinarity reflects a hospital phenotype. While we are unable to elucidate the mechanism of benefit at the high-volume center, it is a real and important observation, nonetheless.

In conclusion, we identified substantial between-hospital variability in donation-related outcomes after OHCA. This patient population also accounted for a quarter of overall organs procured and transplanted in our region. Calls for development or regional systems of post-arrest care and transfer of resuscitated patients to specialty centers have focused on potential to improve survival and functional recovery.^30,31 Our findings parallel these and demonstrate the potential to save additional lives through regionalization of post-arrest care via increased availability of organs for transplantation.